I very much welcome killing CR-only. My execution profiling experience
tells that trying to understand all these options may account
for up to 20% of CPU time in a stateless proxy (depending mainly
on how many characters must be processed which mainly depends
on how messages are built by their senders).

I have not sighted a non-CR-LF implementation, except
debugging messages I send and create on Un*x (LF-only).
(Which is not bad from server's parsing performance perspective 
  ... I always seek LF and ignore optional CR.)

-Jiri

At 05:34 PM 10/5/2001, Jonathan Rosenberg wrote:
>Folks,
>
>My apologies for sending this to sip and sip-implementors, but maximal
>coverage is needed here.
>
>The spec says that implementations MUST send CRLF to terminate lines, but
>MUST interpret CR or LF by themselves as line terminators, for backwards
>compatibility. We want to trim as much of this stuff as possible from the
>spec. So, I ask: is there anyone out there who sends CR or LF alone as line
>terminators? If so, speak up NOW! Otherwise, we will remove this reuqirement
>from the spec and ONLY CRLF will be discussed.
>
>-Jonathan R.

_______________________________________________
Sip-implementors mailing list
[EMAIL PROTECTED]
http://lists.cs.columbia.edu/mailman/listinfo/sip-implementors

Reply via email to